MaskRay added subscribers: urnathan, MaskRay. MaskRay added a comment. I agree that the original patch (c2297544c04764237cedc523083c7be2fb3833d4 <https://reviews.llvm.org/rGc2297544c04764237cedc523083c7be2fb3833d4>) should be reverted.
The `__STDCPP_DEFAULT_NEW_ALIGNMENT__==16` (on x86-64) thing is the glibc behavior. It should be taken with a grain of salt since many malloc implementations don't provide the guarantee for allocations with smaller size (jemalloc/mimalloc/tcmalloc just guarantee 8 alignment on 64-bit systems). And of course, this doesn't apply to to malloc and friends. @urnathan mentioned this on https://github.com/llvm/llvm-project/issues/53540 > Notice the 'suitably-sized allocation' caveat. Either poor wording, or > getNewAlign is not guaranteeing smaller alignment for smaller allocations. I > have not audited all the uses of getNewAlign to see if they presume the value > is a conditional guarantee or not. > GCC also does same assumptions If GCC makes use of `__STDCPP_DEFAULT_NEW_ALIGNMENT__==16` for malloc or operator new with smaller allocations, looks like it is a bug that should be fixed as well. Repository: rG LLVM Github Monorepo CHANGES SINCE LAST ACTION https://reviews.llvm.org/D118804/new/ https://reviews.llvm.org/D118804 _______________________________________________ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
